Will you talk me out of AI if I don't need it?

Yes. If a rule, a script, or a no-code flow does the job cheaper and more reliably, that is what I will recommend. AI only earns its place when it clearly beats the alternative.

Is it safe to put my data through an AI model?

It can be, when it is scoped properly. We agree up front what data the model sees, keep a human in the loop on anything sensitive, and avoid sending more than the task needs. Responsible data handling is part of the design, not an afterthought.

Will AI make decisions on its own?

Only where you want it to. For anything that carries risk, the AI proposes and a human approves. The approval gate is built in, so you stay in control of what actually happens.

All services

AI agents & AI features

AI where it earns its place, not because it's trendy.

Send a message

Most AI projects fail the same way: the model is bolted on because AI is in fashion, not because it solves a real problem. I work the other way round. I start from your process, find the step that actually leaks time or money, and reach for AI only when it clearly beats a rule, a script, or a no-code flow.

When AI is the right answer, it ships like any other production system: a human stays in the loop on anything that matters, inputs and outputs are validated, and your data is handled responsibly rather than piped into a black box. Part of my job is also telling you where AI is just hype and a boring automation would do the job cheaper and more reliably.

What you get

AI used only where it genuinely pays off
A human in the loop on anything that matters
Your data handled responsibly, not piped into a black box
An honest read on where AI is hype and a simpler tool wins

Deliverables

AI agents and assistants wired into your real process
AI-powered steps (classify, extract, draft, route) inside automations
Human-in-the-loop approval gates where they matter
Validation, monitoring, and clear boundaries around the model

Common questions

Will you talk me out of AI if I don't need it?: Yes. If a rule, a script, or a no-code flow does the job cheaper and more reliably, that is what I will recommend. AI only earns its place when it clearly beats the alternative.
Is it safe to put my data through an AI model?: It can be, when it is scoped properly. We agree up front what data the model sees, keep a human in the loop on anything sensitive, and avoid sending more than the task needs. Responsible data handling is part of the design, not an afterthought.
Will AI make decisions on its own?: Only where you want it to. For anything that carries risk, the AI proposes and a human approves. The approval gate is built in, so you stay in control of what actually happens.

Related work

This in the real world.

MARKETING

research → publish, hands-off

Automated content publishing engine

A multi-step pipeline that takes a topic from research to a formatted, published article, with a human approval gate, not a black box.

n8n · Python · Node.js

Case

LEGAL / DOCS

sign & route, automatically

Document generation & e-signing automation

An automation that generates documents from data, routes them for signature, and files the signed copies, removing a manual, error-prone paperwork loop.

Python · FastAPI · PDF tooling

Case

SaaS / CRM

+250% productivity

CRM for teams (Mesoworks)

A CRM with process automation, real-time comms, Make.com integration, and analytics.

Django · Next.js · REST API

Case